It is tempting to rely on heart rate variability (HRV) technology to monitor recovery and predict when we’re ready to train again. Whether it’s an app or a program built into a watch, it’s nice to think that one recovery metric is all we need. But inaccurate recommendations, especially false positives, mean that these programs must be treated with skepticism.
In this article, Scott Semple and Scott Johnston relate their independent experiences with this technology and give you some warnings for using them to monitor your training.
Scott Semple’s Experience
The idea of an app that could tell me when to train was exciting. I liked the convenience of one train-or-don’t-train metric. It would take away the second-guessing and keep me healthy as well. Right?
Unfortunately it didn’t work out that way.
The HRV apps
I started with Elite HRV (because it’s free) and then moved on to ithlete (because it’s nicer to look at). Elite HRV distinguished between sympathetic and parasympathetic biases in the central nervous system. ithlete’s Pro account gave more nuanced prescriptions with its Z-scoring.
But for what I needed—reliable daily training advice—neither was foolproof. Elite HRV recommendations didn’t seem to vary that much. ithlete often told me to train when I felt tired.
Late last year, I did a daily comparison between Elite HRV, ithlete and an orthostatic heart rate test. The apps are quick and simple, but an orthostatic heart rate test is far less convenient. It takes longer and it requires ongoing interpretation, but I had had good results with it in the past. After variable results with the HRV apps, I was willing to try something more cumbersome, as long as it was effective.
In particular, I wanted to look for false positives in any of the three methods. While a false negative—telling me to rest when I could train—could lead to under-training, a false positive would be far worse. False positives—telling me to train when I should rest—would lead to over-training. That would mean lost training time due to excessive fatigue or illness or both.
The test method I used was:
- After waking, I put my on heart rate monitor, lay down, and let my heart rate stabilize; then
- I did the 2.5-minute Elite HRV test; then
- The 55-second ithlete test; then
- The ~4-minute orthostatic test.
The orthostatic heart rate test consisted of:
- Recording my average heart rate over the first two minutes; then
- Standing up and recording my peak heart rate; and finally
- Recording my heart rate one minute after the peak.
Each day I recorded the recommendations of each test method. I added up the “votes” from the tests to come up with a total for the day. As the weeks passed, it became clear what the character of each method was:
- The orthostatic heart rate test (OSHR) was the most conservative. It regularly gave me “red lights”, indicating that I should rest. “Yellow lights” were even more common, suggesting easy recovery days.
- Elite HRV had a more mixed response, not as pessimistic as the OSHR and not as optimistic as ithlete. I like that Elite HRV distinguishes between sympathetic and parasympathetic biases in each reading.
- ithlete was the most aggressive and, therefore, the most concerning. A training readiness test should adopt a do-no-harm policy. ithlete appeared to prefer a train-as-much-as-possible policy.
Two instances in particular confirmed my suspicions about the HRV apps, one with ithlete and one with Elite HRV. In each instance, I was particularly fatigued, and it was obvious that I needed to rest. In contrast, the apps recommended that I train.
Of the two apps, ithlete was definitely the more aggressive in its recommendations. It rarely gave me the red light, while the other two methods were more varied.
I did a lactate test on November 12th where my heart rate reached over 95% of maximum. (My max heart rate is over 200, and I reached 199 during that test.) Three days later, I was still tired, and it felt obvious that I needed a rest day. Both the orthostatic test and Elite HRV agreed.
Not only did ithlete recommend a training day on November 15th, it did even worse. It suggested that my training should be high intensity.
ithlete doesn’t distinguish between biases in the central nervous system. That may explain the bad recommendation. ithlete seems to interpret sharp spikes in variability as enhanced adaptation rather than too much variability. I saw this on several occasions. It then recommends high-intensity training, the worst advice possible.
Unlike ithlete, Elite HRV distinguishes between sympathetic and parasympathetic activity. Each recommendation includes a sign of sympathetic or parasympathetic bias. If Elite HRV sees a sharp spike in HRV, then parasympathetic activity is elevated, and it says it’s time to go easy.
At first, I thought that Elite HRV did a good job of avoiding false positives, but it finally gave one on December 14th.
From December 11th to the 13th, I was on a mountain bike trip. Even before the trip started, I knew I hadn’t planned very well. I was too tired. It was a four-day trip, and we would be riding hard twice a day.
When I woke up on the fourth day, I knew that I had overdone it. I felt horrible, and I worried about getting sick. The orthostatic test and ithlete pointed to a rest day, but Elite HRV recommended training.
This is the “strong like bull, smart like tractor” approach to training: Even when you should rest, keep pushing! I went into this trip tired, and I knew it. The only readiness test that was faithful to how I felt was the orthostatic heart rate test (OSHR). The other two were far too optimistic and, combined with my dumb attitude, dangerous.
The orthostatic test
Why have I assumed that the orthostatic test didn’t produce any false positives? Because whenever I felt like junk, the orthostatic test always raised red flags. It never indicated training when I was excessively fatigued. With “do no harm” as the priority, the orthostatic test was the only test that was faithful to that paradigm.
However, the orthostatic test does have a couple of disadvantages. First, the orthostatic method can take over five minutes while the apps are much shorter.
Second, readings are often uncertain “yellow lights”, even when I think I feel fine. That requires some careful interpretation to make the healthiest choice.
But I don’t see the longer test or the judgment calls as a disadvantage. I’d much prefer to be healthy and under-trained than fall over the edge into extreme fatigue and illness.
“It is remarkable how much long-term advantage [we]have gotten by trying to be consistently not stupid, instead of trying to be very intelligent.”
~ Charlie Munger, Vice Chairman of Berkshire Hathaway
Scott Johnston’s Experience
I started using HRV testing back in 2009. At that time, Polar included Own Optimizer, an HRV recovery tool, in some of their products. Own Optimizer worked well, but it was eventually removed. I heard that this was because customers complained that it took too long.
Since then, I’ve tried several of the HRV apps like ithlete that Scott Semple discussed above. I’ve also used the recent versions of the recovery tools embedded in Garmin and Suunto watches. I’ve tried them on myself and on athletes whom I coach.
It was exciting to think that we could reduce the subjective judgement in coaching. Gauging recovery status is often a tough call for both athletes and coaches. Being able to use hard numbers, rather than feelings, would be a relief.
Between my own and my athletes’, I’ve looked at thousands of workouts over the past eight years. Using that dataset, I’ve been able to form some general observations of HRV technology.
My conclusion aligns with Scott’s. The predictive ability of these devices is far from reliable. Roughly 30% of the training recommendations I’ve seen were in error. Of those, 50% were grossly at odds with the perception of the athlete.
My experience with ithlete was much worse than Scott’s. Using it on myself, it only once told me to rest, on a day that I was utterly spent. On every other day over a four-month period, it told me I was ready to take on the world. That just does not happen.
One young cross country skier I coach used ithlete for a year while training at and near her limit. During that time, it warned her only twice that she might need to rest. Needless to say, those readings didn’t compare well with our observations.
One interesting anecdote that sheds some light on our skepticism came from private communications with the coach of a World Cup cross country skier. This coach was tracking the skier’s HRV using some very high-end software. The data was collected at night while the skier slept and only the coach could view the data. This prevented the athlete’s anxiety from affecting the test results or performance. Overnight tests should provide the most reliability about recovery status. During one particular arduous race week, the software warned that this skier was way into the red zone of fatigue and should rest. That same day she won a World Cup race.
How Can This Be?
The record of this technology among coaches and athletes is depressing. We all want to believe that technology holds the answer and will make the complex simple. A huge amount of peer reviewed science supports HRV as it relates to stress. The nervous system should be the first system in the body to react to stress. So, why can’t this technology live up to the promises made for it?
I have no proof of my pet theory other than 40-plus years of being an athlete and coach combined with thousands of hours watching how I and others adapt to training. Here is my take on it:
We are incredibly complex organisms. Our many systems interact in ways we don’t fully comprehend. I have no doubt that these HRV measurements are accurate. But we can’t rely upon a single metric to describe the state of the whole organism. We are too complex for such a simple model to offer reliable predictive abilities.
HRV technology seems great at back-casting—looking at a past outcome and showing why it occurred. But for what serious athletes need—reliable, daily advice on if and how hard to train in the immediate future—it falls short.
We don’t see any benefit to using current HRV apps. In fact, telling athletes to train when they should rest is potentially harmful. We hope to get better results in the future.